The Research On Multi-objective Virtual Network Embedding Algorithm

In recent years,the Internet as the underlying infrastructure is also under-going significant changes.In order to better guarantee a variety of business innovation and network services over the Internet,researchers have proposed network virtualization technology to fend off this ossification dilemma.Lever-aging network virtualization technology,various computing,network,storage and spectrum resources in the underlying substrate network infrastructure are managed in an unified manner,which can support on-demand manner to pro-vide services for various end users.Virtual network embedding issue is one of the core issues in network virtu-alization technologies.The main purpose is to optimize the allocation algorithm of various resources as much as possible under the premise of limited substrate network resources,and to maximize the utilization of substrate network re-sources,with the aim of satisfying the requirements of more virtual network requests(VNRs).This thesis focuses on the model building of virtual network embedding problem,energy consumption awareness embedding algorithm,se-curity awareness embedding algorithm,and service awareness embedding al-gorithm.On this basis,a multi-objective virtual network embedding algorithm considering energy consumption,load balancing and revenue is proposed.The main contributions of this thesis are summarized as follows.1.The thesis devised a virtual network embedding algorithm model based on three-dimensional resource constraints.Since scholars generally agree with the principle of "storage for band-width",it is particularly important to consider the dimension of "storage" in virtual network embedding algorithm.In this thesis,"storage" is added to the virtual network embedding algorithm model for the first time,and a virtual net-work embedding algorithm model based on computing,network and storage three-dimensional resource constraints is proposed.With the help of heuristic algorithm,two benchmark algorithms are devised.A large number of simula-tion experiments demonstrate that the proposed algorithms are feasible and ef-fective.The proposed two benchmark algorithms can be used as the basis for fu-ture research on virtual network embedding algorithms with multi-dimensional resource constraints.2.This thesis proposed an energy-aware virtual network embedding algo-rithm using niche particle swarm optimization.In the environment of data center network,with the increasing demand for green environment protection,energy consumption account for an important part of its operational cost.In this thesis,a multi-objective enhanced virtual network embedding algorithm based on particle swarm optimization(PSO)is proposed.Fitness function is designed using aggregation strategy,and niche particle swarm optimization(niche PSO)is used to avoid falling into local op-timum.The simulation results show that the algorithm has obvious advantages in the comprehensive performance of energy consumption and revenue.3.This thesis proposed a secure virtual network embedding algorithm us-ing information entropy TOPSIS method.Internet security incidents occur frequently,Internet users pay more and more attention to the security problem.The thesis presents a security-aware virtual network embedding algorithm,which objectively evaluates the impor-tance of substrate nodes using information entropy TOPSIS method,and choos-es more suitable substrate nodes for virtual nodes in VNRs.Experiments show that the proposed algorithm has better performance than the traditional security-aware virtual network embedding algorithms.4.This thesis proposed a service-aware virtual network embedding algo-rithm based on the reliability of nodes and links.With the emergence of various new businesses over the Internet,the future trend must be that all kinds of users have a variety of service requirements.From the point of view of the reliability of substrate nodes and substrate links,under the condition of limited substrate network resources,this thesis uses the reliability measurement parameters of nodes and links to allocate more reliable substrate network resources for VNRs,with the aim of improving the quality of service of VNRs.5.This thesis proposed a virtual network embedding algorithm based on multi-objectives of energy consumption,load balancing and revenue.Different users have different quality of service(QoS)requirements,and there are conflicts among different optimization objectives.Designing multi-objective optimization virtual network embedding algorithm could satisfy the QoS requirements of most users.This thesis presents a virtual network em-bedding algorithm based on energy consumption,load balancing and revenue optimization objectives.In the node embedding phase,the available resources,load balancing and energy consumption parameters of the nodes are incorpo-rated into the importance evaluation process of the substrate nodes.In the link embedding phase,the improved differential pricing strategy is utilized to realize the substrate path selection process.Finally,we summarized the main work and innovations,and outlined the future research work.